react-native-executorch

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill downloads and processes arbitrary remote resources: ResourceFetcher.fetch and model loading accept 'https://...' model URLs, and multiple hooks (useClassification/useOCR/useImageEmbeddings/useVAD/useSpeechToText and examples using 'https://url-to-image.jpg' or 'https://some-audio-url.com/file.mp3') consume and interpret untrusted public URLs at runtime, which can enable indirect prompt injection.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.70). The skill demonstrates runtime downloading/loading of model binaries from external URLs (e.g. Hugging Face models at https://huggingface.co/software-mansion/collections) via ResourceFetcher / modelSource, which are required for operation and result in executing remote model artifacts locally.